GUI for Color-Blindness

We have a need to display values in a GUI "waterfall-type" display, with ti me on the Y-axis and frequency on the X-axis. I'm wondering about whether or not we should add an option button to display an alternate version to as sist those with color-blindness. In short, I'm wondering if this is a big deal or not?

We will be displaying values from -120 dBm to -90 dBm. "-120" is good. "-90" is bad. There is also the possiblity of "No Data", which I'm thinking should be eit her white, or black. There is also the remote possibilty of buffer overflo w, which isn't the same thing as "No Data". That case should be extremely rare, but we can't rule it out.

So my questions:

1) Is there a recommended methodology to assign colors to the discrete valu es (either individually, or in blocks of say... 5 dBm)? My present thinkin g is each dBm value gets assigned a unique color. My hope is for a nice, pleasing display - since this output will mostly be viewed by non-engineers . (who can see better than they can think). OK, that comment will probabl y get me in trouble if read by the wrong folks! :) But, you get my point. Hopefully.

2) Are there standards for this? i.e., color-coding of values to "maximiz e" reader comprehension and retention?

3) For both 1) and 2) above, I need (may need?) to do the same thing to fa cilitate those with color-blindness - which I'll limit here to red-green co lor blindness, which as I understand it, is supposedly the most common type .

I could spend a few hours fiddling with it to see what "looks good" to me, but would rather just have the RGB hex values and be done with it.

Any ideas or suggestions? Thanks.

Reply to
mpm
Loading thread data ...

Something like 8% of the white male population has some kind of colorblindness.

Looking at the "difficult" combinations here:

formatting link
your-blind-colors/

it seems that something like orange on dark blue for an x-y display would work for most variations of color blindness...

----Android NewsGroup Reader----

formatting link

Reply to
bitrex

Or just switch the display to greyscale, if it looks good to you it will probably look good to them?

--


----Android NewsGroup Reader---- 
http://usenet.sinaapp.com/
Reply to
bitrex

You know, I just looked at that site and, wow. I never knew there were so many types.

You can either test all of them and tailor the display for the group, or risk getting charged by the ADA. That's right, if anything qualifies as disability in something like this, this does.

Think you can do it in greyscale ? Only other thing is just whatever parameter is, umm, dunno the word, up front is at one oend of the visual spectrum and the rest head toward the other ? Red, orange, yellow, you know.

The only ways I can "see" to do it.

Reply to
jurb6006

HA, too bad you can't rename the thread.

"Daltonism Is A Disability"

Reply to
jurb6006

This kind of depends on what kind of screen it will be on and what the ambient lighting will be like. Also, is it important for the user to tell -110 dBm from -112 dBm, or do they only need to tell -110 dBm from

-100 dBm?

If that leads you to 30 colors, my opinion is that's probably OK. If you decide you want fewer colors, I'd say to try every 2 dBm, for 15 colors.

Probably, but I don't know what they are. :) There are "obvious" ones like green=good, red=bad, or blue=cold, red=hot, but I'm not sure if there is a standard you can reference. If there is a user interface / user experience designer in the building, ask him or her.

The rule of thumb I've heard for this is that for things like indicators that are either off or on, something other than the color should also change when the indicator changes state. Like, drawing a (black) border around it or not, or having a text display of "off" or "on", or whatever. But this sort of falls down for a more complicated display, like you have.

There are software tools (local and online) that will show you what a given image looks like for various kinds of colorblindness, and/or adjust your desktop display to simulate it. A quick Google gives:

formatting link
formatting link
formatting link

One thing I've seen done, in cases where you don't want to give the user the ability to edit the color of every single thing on the screen, is to have, say, 3 to 5 predefined color schemes, and a button that says "change color scheme", which just rotates through all of them in order. I am not color-blind and usually I find one scheme looks pretty good, a couple look OK, and a couple look hideous - I assume the hideous ones look OK to people who see differently than me.

You can also prototype it in Excel (or equal) where it's easy to change the colors, show several different color schemes around the office, and ask for people's opinions.

Matt Roberds

Reply to
mroberds

I think the term you want to Google for is "color map". The work to let your user choose from half a dozen is only slightly more of a step than the difference between only allowing one and allowing two, and you don't have to have a "color blind" option (which some won't want to use because they don't want to be noticed, and some won't want to use because they have no clue they're color blind).

If you offer a number of different color maps then you'll cover color blindness of all sorts, lighting issues, personal preference, and a whole lot of other nifty things.

--
www.wescottdesign.com
Reply to
Tim Wescott

Potentially yes since about 8% of males are colour blind to some extent. Although a listbox is probably more suitable for this purpose.

The false colour maps that map onto monotonically increasing luminance and avoid one of red or green are the most nearly failsafe.

That is shades of grey, shades of a single colour or the thermal map, or a classical intensity one from astronomy which although it includes some forbidden colours seems to work OK for most people. Ironbow in thermal imaging is another popular choice. My personal favourite false colour map is a tweak of Peano0 which traces around the edges of the colour cube with a twist to allow it to start at black and end at white.

Spectrum is possible but again folk some will not see red/green or blue/yellow as you intend. Providing a decent range of false colour palettes is the usual way around this including the monochrome.

A selection of my favourite false colour maps I have collected along the years is online at:

formatting link

Various other scientific disciplines have their own particular favourites but the most common ones are all there. I should rework them sometime using more recent just noticeable colour difference methods since there are too many shades of green and not enough shades of yellow or cyan for maximum discrimination with normal colour vision.

Reminds me that I must get around to updating it... roundtuits needed.

Usual method is to map the data range onto a colour palette of 256 values with any special reserved codes at the top end.

If you only need 30 then a simple greyscale will be good enough. Few monitors are so badly adjusted they can't distinguish 30 shades of grey! False colour makes it a lot easier to see features though.

But a false colour map applied to a photograph can be a disaster!

Not standards exactly but various false colour palettes are in common use in some form. Most domains have their own particular favourite which emphasises whatever data features are most likely to occur.

--
Regards, 
Martin Brown
Reply to
Martin Brown

I second the greyscale recommendation. There is no "intuitively natural" ordering of colours.

Reply to
Ralph Barone

You should add a "calibration scale" close to the actual data showing different greyscale for 0..100 % or xx .. yy dBm.

Reply to
upsidedown

In those two special cases, all frequencies are bad, so you could use diagonal lines for those areas, e.g. top left to bottom right = No Data and top right to bottom left = Overflow.

Reply to
upsidedown

My former employer, a Physicist, was is colorblind. He prefers greyscale for imaging work.

If you have a wide range of data, do what is done with medical imaging. A pair of sliders is moved with the mouse to adjust the Gamma Curve and Offset.

The slider technique is essential with CT and MRI imaging.

When offered false color, most Radiologists will insist on going to greyscale.

I learned that the hard way when repairing CT machines.

Steve

Reply to
sroberts6328

So I go and look and the sliders are called Window Width and Window Offset.

Here is a decent explanation of how you display a range of over 4000 units on a 256 unit display:

formatting link

Steve

Reply to
sroberts6328

Thanks! I am fast coming to that realization.

Also, thanks everyone else for the help.

Reply to
mpm

False colour can make difficult complex data more easily interpretable at a glance but it is often a total disaster for greyscale images.

We tended to use default false colour palette that with a just flick of the switch would render correctly in monochrome luminance only.

The other one which is potentially useful is a greedy first histogram equalisation that maps the original large dynamic range image onto the

256 bins available in such a way as to make them contain 1/N th or more of the remaining total values. A maximum entropy solution to preserve info and make fine gradations more obvious. It can be very good at making tiny systematic errors visible to an unskilled practitioner.

You can do a lot with linear maps, curves and gamma.

--
Regards, 
Martin Brown
Reply to
Martin Brown

Don't thank me. Thank Edward Tufte.

Reply to
Ralph Barone

The various "High Contrast" themes, whether in color or monochrome, are easy to discern between elements.

Reply to
DecadentLinuxUserNumeroUno

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.